Treadmill exercise capacity and hemodynamic, symptomatic, and electrocardiographic (ECG) responses to leg exercise are powerful predictors of all-cause mortality and other important measures of health outcome but none of this information is available for patients who undergo pharmacologic stress tests because of lower extremity disabilities such as amputations, knee or other arthropathies, myopathies, claudication, spinal cord, or combat injuries. The major hypothesis of this project is that composite findings of arm ergometer exercise ECG stress testing, with selective myocardial perfusion imaging (MPI) in a subset of patients such as those with highly abnormal resting ECGs or blunted hemodynamic responses, provides prognostic, clinical, and diagnostic information equivalent to non-selective pharmacologic MPI at a fraction of the cost, time expenditure, radiation exposure, and institutional equipment and personnel resource requirements in patients who cannot perform adequate lower extremity exercise. To investigate this hypothesis, we propose the following specific aims and objectives: Aim # 1 Objective # 1 is to compile data on demographics, baseline clinical characteristics, medications, stress test and MPI responses to stress, and subsequent clinical outcome in a computerized research database obtained from electronic records of approximately 5800 unique veterans who underwent 360 arm ergometer, over 2000 dipyridamole, and 3400 treadmill stress tests performed on patients at the St. Louis Veterans Administration Medical Center between 1997 and 2002. Objective # 2 is to determine by univariate and multivariate Cox regression analyses, Kaplan-Meier curves, and Wald G2 models, which exercise capacity, hemodynamic, symptomatic, ECG and MPI responses to arm ergometer, dipyridamole and treadmill stress tests independently predict survival or all-cause mortality, subsequent myocardial infarction (MI), or coronary revascularization alone and in combination, after adjustment for demographic, clinical, and medication variables, during follow-up to an endpoint of death or 12/31/2009. Objective # 3 is to compare the prognostic value of a model of independently predictive arm ergometer ECG stress test variables with dipyridamole MPI, treadmill ECG, arm ergometer and treadmill ECG plus MPI results, using receiver operator characteristic (ROC) curve analyses, for prediction of clinical outcome. Aim # 2 The goal is to develop a simple scoring system, analogous to the Duke Treadmill Score, to best prognosticate death and MI from independently predictive arm ergometer ECG stress test variables. Aim # 3 Objective # 1 is to stratify subgroups of patients in the above data base with a highly abnormal baseline ECG, blunted hemodynamic responses to stress, an abnormal stress ECG or low exercise capacity, for whom it is hypothesized that MPI information might be most beneficial for incremental prognostication of outcome. Objective # 2 is to investigate the independent incremental prognostic value of MPI in these subgroups and compare the incremental value of MPI for prediction of outcome in the various subgroups of veterans using Cox regression and G2 analyses.